Create gd.py

Author: Mayurji

Day-23-Gradient-Descent/gd.py

@@ -0,0 +1,58 @@
+import torch
+from torch import nn
+from sklearn.datasets import load_boston
+
+class GradientDescent:
+    def __init__(self, learning_rate=0.01, max_iterations=100):
+        self.lr = learning_rate
+        self.max_iterations = max_iterations
+
+    def normalization(self, X):
+        """
+        :param X: Input tensor
+        :return: Normalized input using l2 norm.
+        """
+        l2 = torch.norm(X, p=2, dim=-1)
+        l2[l2 == 0] = 1
+        return X / l2.unsqueeze(1)
+
+    def compute_error(self, b, m, X, y):
+        total_error = 0
+        for i in range(0, X.shape[0]):
+            total_error += (y - (torch.mm(m , X.T)) + b) ** 2
+        return total_error / float(X.shape[0])
+
+    def step(self, b_curr, m_curr, X, y, learning_rate):
+        b_gradient = 0
+        m_gradient = 0
+        N = float(X.shape[0])
+        for i in range(X.shape[0]):
+            b_gradient += -(2/N) * torch.sum(y - (torch.mm(X, m_curr.T) + b_curr), dim=0)
+            m_gradient += -(2/N) * torch.sum(torch.mm(X.T,  (y - (torch.mm(X, m_curr.T) + b_curr))), dim=0)
+
+        new_b = b_curr - (learning_rate * b_gradient)
+        new_m = m_curr - (learning_rate * m_gradient)
+        return [new_b, new_m]
+
+    def gradient_descent(self, X, y, start_b, start_m):
+        b = start_b
+        m = start_m
+        for i in range(self.max_iterations):
+            b, m = self.step(b_curr=b, m_curr=m, X=X, y=y, learning_rate=self.lr)
+
+        return b, m
+
+if __name__ == '__main__':
+    data = load_boston()
+    X = torch.tensor(data.data)
+    y = torch.tensor(data.target).unsqueeze(1)
+    initial_b = 0.0
+    initial_m = torch.zeros((X.shape[1], 1), dtype=torch.double).T
+    nn.init.normal(initial_m)
+    gd = GradientDescent(learning_rate=0.0001,max_iterations=100)
+    gd.compute_error(X=gd.normalization(X), y=y, b=initial_b, m=initial_m)
+    bias, slope = gd.gradient_descent(gd.normalization(X), y, start_b=initial_b, start_m=initial_m)
+    X = gd.normalization(X)
+    print('y: ', y[0].item())
+    print('y_pred: ', (torch.mm(slope, X[0].unsqueeze(0).T)+bias).item())
+